Confounding means an effect of interest is entangled with something else. In designed experiments it is often intentional: a high-order interaction is confounded with blocks so the block effect costs nothing that matters. In observational data it is the main trap: if all new-recipe lots also ran on the newest tool, any yield gain is confounded between recipe and tool, and no amount of analysis will pull them apart.
The new slurry went only on CMP 3, so the defect gain is confounded with the tool. We need a proper split.
Also heard as
- confounded
Related terms
Aliasing
Situation in a fractional design where two or more effects leave identical patterns in the data and cannot be told apart.
Block
Group of experimental runs made under similar conditions, used to remove a known nuisance source such as day, batch or tool.
Randomization
Running experimental trials in a random order so unknown time trends and nuisance effects do not bias the results.
Split lot
Lot of wafers divided between two or more process conditions so the comparison shares the same incoming material.